In Situ representations and access consciousness in neural blackboard or workspace architectures

Abstract

Phenomenal theories of consciousness assert that consciousness is based on specific neural correlates in the brain, which can be separated from all cognitive functions we can perform. If so, the search for robot consciousness seems to be doomed. By contrast, theories of functional or access consciousness assert that consciousness can be studied only with forms of cognitive access, given by cognitive processes. Consequently, consciousness and cognitive access cannot be fully dissociated. Here, the global features of cognitive access of consciousness are discussed based on neural blackboard or (global) workspace architectures, combined with content addressable or "in situ" representations as found in the brain. These representations allow continuous cognitive access in the form of a process of covert or overt queries and answers that could underlie forms of access consciousness. A crucial aspect of this process is that it is controlled by the activity of the in situ representations themselves and the relations they can initiate, not by an external controller like a CPU that runs a particular program. Although the resulting process of access consciousness is indeed based on specific features of the brain, there are no principled reasons to assume that this process cannot be achieved in robots either.